Introduction
I’ve assembled a collection of playlists reflecting my music preferences, with a focus on KRnB and KPOP genres. These playlists fall into four categories: two are custom-tailored to my individual tastes in KRnB and KPOP, while the other two offer a broader selection of KRnB and KPOP tracks.
The personal playlists are curated by Spotify, leveraging my listening history to deliver content tailored to my preferences. Conversely, the broader KRnB and KPOP playlists appeal to a wider audience.
Spotify-curated playlists already contained 50 songs each, while for the broader playlists, I randomly picked 50 songs from larger playlists.
With this collection, I aim to analyze and understand the differences between KRnB and KPOP. I’m also interested in exploring why I prefer certain KPOP and KRnB tracks over others. This analysis will help me gain insights into my music preferences and the unique characteristics of these genres.
Portoflio
First i want to use clustering to show whether significant differences exist between the genres of KR&B and K-pop, while also providing an overview of the prevalent features within each genre.
After this I delved deeper into the individual features of the songs
And finally the conclusion
Personal Playlists
General Playlists
Complete Linkage Dendogram
In this dendrogram, I picked out 10 K-pop tunes and 10 K-R&B tracks from my collection. When we look at the complete linkage dendrogram, we can spot a clear moment where the dendrogram splits into three clusters. I marked this with a vertical red line. Taking a closer look, one cluster mostly has K-R&B songs, another is dominated by K-pop tunes, and the third cluster has only K-R&B tracks.
From this, it seems like hierarchical clustering does a decent job of sorting songs in my collection by genre. But it’s not perfect. Sometimes, it mixes songs from different genres, indicating that it’s not flawless.
This may be due to the subjective nature of categorizing songs into genres, even leading to debates about which genre a song belongs to.
Heatmap
This heatmap displays the complete linkage dendrogram for the songs, alongside one for the feature values. It’s fascinating to observe the separation of the tempo and chroma values and the timbre values into two distinct clusters. One intriguing observation is the outlier status of “New Jeans” in terms of instrumentalness. However, upon listening to the song and comparing it to, for example, “Hola,” it seems that the latter should have a higher instrumentalness value.
K-Nearest Neighbour
The KNN (K-Nearest Neighbors) classifier achieved moderate performance in classifying songs into K-pop and K-R&B genres. With precision scores of approximately 0.71 for K-pop and 0.64 for K-R&B, the model correctly classified a significant portion of instances. However, the recall scores were lower, indicating that some genre instances were missed. Notably, K-R&B had a higher recall but lower precision compared to K-pop.
Tempo Histogram
As you can see from the histogram, most K-pop songs tend to hover around 125 beats per minute (BPM), while KR&B songs often have a spike around 85 BPM. However, it’s common for songs in both genres to stray from these averages. One standout outlier among KR&B tracks is “FLYING HIGH WITH U” by Vinxen. On the other hand, “shopper” by IU is one of the K-pop songs with a lower BPM. You can find these outliers by hovering over the plot, of which the tooltip shows the title and the BPM of each song.
Now, let’s take a closer look at the tempogram analysis of these songs.
Shopper - IU
The tempogram displayed here is a non-cyclical one, meaning it may show multiples of the main BPM. In this case, it’s interesting because Spotify tagged the song with a BPM of 75, but the tempogram indicates a BPM of around 200 for most of the song, with only a small dot around 75 BPM at the beginning. However, when you listen to the song, it starts to make sense. At the beginning, there’s a section with a distinct clapping noise that seems to follow a rhythm of 75 BPM, uninterrupted by other instruments except for the vocals. This is followed by a beat that might sound like 200 BPM but is actually closer to 75 BPM to the human ear.
FLYING HIGH WITH U - Vinxen
Interestingly, the tempogram for this song does align with the BPM provided by Spotify, which is around 210 BPM. The tempogram even indicates activity for multiples of 210 BPM. However, despite this numerical match, when you actually listen to the song, it doesn’t sound like it has such a fast tempo. This discrepancy might be due to the difference in tempo between the drums and the vocals. While the drums might be driving the high BPM detected by the tempogram, the vocals maintain a slower pace, resulting in an overall perception of a slower tempo.
Key Histogram
In the genre of KPOP, the most prevalent keys are C sharp, F, and G, as indicated by the plots. Conversely, in KRnB, the histogram reveals that the A and G key is favored among other keys.
In KPOP, the prevalence of C sharp, F, and G keys mirrors their common usage in Western pop music, suggesting a shared musical influence between the two genres. Conversely, within KRnB, the dominance of the A and G keys underscores the unique tonal preferences characteristic of Korean R&B music. These trends underscore the dynamic interplay between cultural influences and genre-specific characteristics, illustrating how musical styles can both converge and diverge across different contexts.
Keygrams for songs in different keys
Now lets look at the keygrams.
In looking at the keygrams of the most popular songs in the corpus, some interesting patterns emerge regarding their tonal centers.
Song in the key of D
For instance, when looking at the song in the key of D, we notice that the confidence for D minor key dominates, which suggests a strong association with that tonality. But what’s really interesting is that the confidence for C sharp minor key pops up even more frequently than for D minor. This might be because of some connection between D minor and C sharp minor keys. The keys share some of the same notes, which could explain why C sharp minor shows up so prominently alongside D minor.
Song in the key of G
Now, the keygram of the song in G key is a bit more straightforward. It’s clear that both G minor and major keys rule the roost throughout the song, indicating a strong focus on G as the tonal center. Also interesting to note: there are a couple of moments in the song where the keygram isn’t too sure about which key it’s in, besides the usual uncertainty at the beginning and end of songs. These moments of uncertainty could signal shifts in harmony or brief changes in tonality within the song.
I’ve chosen two songs, using trial-and-error to see which one has a clear structure and which one doesn’t. The song with the clear structure is called Cookie by NewJeans and the song with the less clear structure is called 21 by DEAN.
Clear Structure
With over 200 million streams on Spotify, “Cookie” stands out as one of the most popular K-pop songs of its generation. With such widespread popularity, it’s no surprise that the song boasts a clear structure, evident even in its timbre plot.
Take a look at the plot, and you’ll notice distinct blocks and paths. These blocks represent homogeneous sections, like verses or choruses. Understanding that the song follows a Verse-Chorus-Verse-Chorus-Bridge-Chorus structure helps decipher these blocks and paths.
Each block corresponds to a verse or a chorus, making it easy to distinguish between different sections of the song. Furthermore, as you listen to the song, you’ll likely pick up on how variations in vocal timbre contribute to the distinct blocks seen in the timbre plot.
Additionally, the path-like structures that run parallel to the main diagonal indicate repeated sections. In “Cookie,” these path-like structures prominently appear in each chorus, emphasizing the song’s repetitive elements.
Unclear Structure
“21” by DEAN, boasting over 30 million streams, is undeniably popular, following a verse-chorus structure like many hits. Despite its popularity, the song’s timbre remains relatively uniform throughout, making it challenging to discern distinct block structures. However, the path-like structures characteristic of a chorus are still noticeable, guiding listeners through the song’s familiar sections.
Interestingly, examining the chroma plot reveals the bridge section, a detail not as evident in the self-similarity matrix of “Cookie” by NewJeans.
Chromagram Comparison
I selected two songs from the corpus based on their energy and loudness scores provided by Spotify. One of the songs is “LIMBO” by keshi, which scores low in both energy and loudness. The other song is “Talk That Talk” by TWICE, which scores high in both energy and loudness.
When comparing the chromagrams of these songs, the differences are quite apparent. “Talk That Talk” is notably higher in tempo compared to “LIMBO.”
In terms of musical composition, “Talk That Talk” employs a wider range of notes and chords whereas “LIMBO” maintains a simpler structure with a clear emphasis on the G sharp/A flat note.
Regarding the data from the Spotify API, it seems less confident in identifying the notes used in “Talk That Talk” compared to “LIMBO.” This could be attributed to the greater variability in the composition of “Talk That Talk.”
Overall, these observations underscore how songs can differ significantly in their energy, and musical complexity, as perceived by listeners.
# A tibble: 1 × 62
playlist_id playlist_name playlist_img playlist_owner_name playlist_owner_id
<chr> <chr> <chr> <chr> <chr>
1 2TJHOlrTUoLM… The Sound of… https://ima… The Sounds of Spot… thesoundsofspoti…
# ℹ 57 more variables: danceability <dbl>, energy <dbl>, key <int>,
# loudness <dbl>, mode <int>, speechiness <dbl>, acousticness <dbl>,
# instrumentalness <dbl>, liveness <dbl>, valence <dbl>, tempo <dbl>,
# track.id <chr>, analysis_url <chr>, time_signature <int>, added_at <chr>,
# is_local <lgl>, primary_color <lgl>, added_by.href <chr>,
# added_by.id <chr>, added_by.type <chr>, added_by.uri <chr>,
# added_by.external_urls.spotify <chr>, track.artists <list>, …
Energy Distribution
In this plot I pulled in data of 4 different playlists:
Let’s first look at the personal playlists by selecting that option in the plot legend.
What we see is that, on average, KPOP songs tend to have higher energy levels compared to KRnB. But there’s a catch – there’s some overlap between the two genres. This means that a KRnB track could have the energy level typical of a KPOP song, and vice versa. It goes to show that energy level is just one factor among many that define a song’s genre.
Now, when we switch over to the general playlists, things look different. The songs seem more spread out across energy levels compared to the personal playlists. This suggests that personally, I might have a specific energy level in mind when picking KRnB or KPOP songs for my playlists. It’s all about personal taste and what vibes with me the most.
Acousticness Comparison
Looking at the plot, it’s clear that KRnB songs tend to have higher acousticness compared to KPOP tunes. There’s just a small sliver of overlap between the acousticness of KRnB and KPOP tracks. Also, notice how the range of acousticness for KRnB songs is wider than that of KPOP. This suggests that KRnB artists have more leeway in using different instruments compared to the more structured sound of KPOP.
In the plot, you can clearly see how energy and loudness are linked, especially with that regression line slicing through the data points. There’s also this faint boundary line that seems to separate KPOP and KRnB songs. Oh, and there’s this one standout: the KPOP hit “Cupid” by Fifty Fifty, chilling in the KRnB zone for loudness and energy. Just hover over the point to check it out!
After thorough exploration, it’s evident that KRnB and KPOP are distinct genres with noticeable differences. KPOP tends to have a higher overall tempo compared to KRnB. Moreover, the choice of keys varies significantly between these genres.
In terms of song structure, KPOP tends to exhibit more discernible Self-Similarity Matrices (SSMs), indicating clearer patterns. Additionally, when examining chromatic differences, KPOP tends to utilize a wider range of chroma.
Furthermore, KPOP generally exudes more energy and features less acoustic elements compared to KRnB. This difference in energy aligns with my personal taste, which tends to lean towards more specific preferences across various features.
In summary, while both KRnB and KPOP offer unique musical experiences, their distinct characteristics appeal to different tastes and preferences.